Multimedia
Blogs
Forum
Affordable HD
Whitepapers
Newsletters
Events
Advertisers
DCP Directory
MillimeterHDV editing using Avid’s Liquid 7.1, Part 3
Aug 14, 2006 11:27 AM, Steve Mullen
In the first installment, I asked “What would it mean to your HDV editing workflow if you could insert multiple video formats into a timeline?” We then began to examine this question by looking at Avid’s Liquid 7.1, with which I had created two HDV sequences. (See Part 2 here.)
One sequence was HDV 720p30 and the other sequence was HDV 1080i60. The 720p30 sequence contained 720p30, 720p24, 1080i60, 16:9 DV, 4:3 DV, and QCIF MPEG-1. The 1080i60 sequence began with multiple 1080i clips followed by the entire 720p sequence. In this installment, we will look at how I created these sequences.
For those of you who have edited DV exclusively, the obvious question is what kind of sequence can hold multiple formats. You have learned, often the hard way, that bad things happen when your sequence preset does not exactly match your DV source in every detail. Obviously, this rule cannot be applicable to an NLE that supports multiple formats in a sequence.
To understand why this rule no longer holds, I’ll ask a rhetorical question. Why do we need to define the nature of a sequence? The answer is that the NLE needs, at minimum, two items of information: image aspect ratio and pixel aspect ratio. The first is needed to define the aspect ratio of the recorder monitor created on our computer’s monitor. Note, the player monitor’s aspect ratio(s) is defined by the source format(s). Pixel aspect ratio is required when conversions are made between square and non-square pixel video.
You will note that I did not define frame rate to be one of the necessary sequence parameters — and it’s not. It is possible for a timeline to be defined in terms of time. In such a timeline, a second is a second. Liquid operates using this system. The playback engine notes a source’s frame rate and displays frames at exactly the rate necessary for one second’s worth of video to be presented in one second. Likewise, a source’s audio sampling rate determines how many samples must be output per second. This is how multiple frame and sample rates can be mixed within a sequence.
Unfortunately, Liquid does not take this capability to its logical conclusion—a timeline where timecode is fully replaced by “time.” With time notation, a gap from 5.25 minutes to 8.50 minutes would require 3.25 minutes of source. It would make no difference what frame rate were used by the source.
With this understanding, let’s create a new sequence in Liquid. We can give the sequence a unique name — or let Liquid auto-increment a name. We can then select a Layout Template. (I will cover this time-saving feature at the end of this Installment.)
Now, if we like, we can choose from a list of presets that currently includes this subset of DV and HDV that's relevant to Region 60: NTSC 4:3, NTSC 16:9, HDV 720/29.97p, and HDV 1080/59.94i. Obviously, if I were working with Sony HDV, I would simply select “HDV 1080/59.94i.” This would automatically define frame rate, resolution, interlace field order, and M2V codec.
When a frame of video is needed by the NLE’s display or effects engine, the appropriate codec is employed to decompress (DVC) or decode (MPEG) to uncompressed YUV data. For display on a computer monitor, the YUV data must be converted to RGB data. However, if your computer has a board that outputs analog component or HD-SDI, the data are not converted and so remain in YUV colorspace.
Depending which NLE you use, its effects engine works either in YUV or RGB colorspace. (One exception is Premiere when using CineForm YUV effects rather than Premiere’s RGB effects.) There are two advantages of working with only with YUV data. First, color errors can result from the conversion to and from YUV and RGB colorspaces. Second, video clipping can occur when YUV data are converted to RGB data.
A recorded video signal often is higher than 100IRE — yielding digital YUV values greater than “240.” (Legal digital values range from “16” for 0IRE to “240” for 100IRE.) Moreover, if you alter your camera’s Pedestal setting, recorded data may be lower than “16.” When YUV data are converted to RGB data, the YUV values between “16” and “240” are mapped to RGB values from “0” to 255.” Thus, YUV values below “16” or above “240” are truncated and valuable information is lost. (With FCP, you must enable “Super White” to prevent truncation.)
While Liquid’s realtime effects work in RGB colorspace, you can avoid clipping by adjusting White and/or Black levels using the Color Corrector. Liquid has several CC functions — one that supports both Primary (realtime) and Secondary (non-realtime) correction as well as a vectorscope and waveform monitor. (This CC has White and Black level controls as well as a Soft Clip function.) Unfortunately, this corrector does not support key-frames. The alternate, Base, CC function does support key-frames — although regrettably, only realtime Primary correction is supported and scopes are not available. Both correctors map YUV 0-255 to RGB colorspace that employs 16-bits per-component. Thus, no input signal truncation occurs when using these correctors.
Tip 5: For work that must meet broadcast standards, Liquid’s Soft Clip function, that limits video to 100IRE, should be applied to each clip. The Soft Clipping feature is found within Liquid’s Color Correction Legalizer. While you must enable the CCIR-601 function, you should not enable the “Work only in CCIR-601” mode. Now, while watching Liquid’s waveform monitor, dial in the amount of soft clipping you want. You should also enable the “75% color” function that limits color intensity.
The three 16-bit components are scaled to 24-bits for display during correction — as well as after processing. If the processed data are fed to an RGB realtime effect, the signal below 0IRE and above 100IRE will be truncated. However, when sent to a YUV-based effect or compressed, encoded, written as uncompressed video to a file, or output — no truncation will be performed. Color correcting HD naturally brings up the question of how you can monitor your efforts. I have connected the DVI output from my Dell laptop to the HDMI input of my Sony 1280x720, LCD HDTV. Now either Liquid’s Player or Recorder window is displayed on the HDTV. Naturally, you will need to calibrate an HDTV to obtain an accurate image. (In fact, any monitor should be calibrated.)
Liquid’s “VGA Second Head” option outputs video in RGB colorspace. Whenever RGB data are viewed, there may be slight color differences between the display and the YUV video. However, since many folks editing HDV will be using consumer HDTVs for monitoring, I’m not convinced the colorspace error is a serious issue for them.
Tip 6: The Liquid Second Head function does not use a VGA driver’s “overlay” capability. Rather, it uses the VGA driver’s second monitor capability. (Do not enable the “Desktop across both monitors” mode.) Under Edit > Control Panel > Site > Video Outputs > VGA Second Head — define “use monitor setup” and “custom framerate = 60.” Whether you use your HDTV’s DVI/HDMI or RGB input, you’ll have to discover a setting for your PC’s graphics driver that fills, as much as possible, your monitor’s screen. This is a topic for another day.
Templates provide efficient ways of working on different projects. For example, I’ve defined 24p and 30p 720 HDV as well as 1080i HDV master sequences. Each sequence includes leader, bars, tone, and trailer. After defining these sequences, I saved them as templates. (I defined one as the Default Template.) Now, when I create a new project sequence — one that will contain the final sequences—I simply choose the appropriate template. The new sequence is immediately ready for use.
In the next installment, I will look at the exact role codec choice plays in a sequence preset. We will confront one of most fundamental debates in working with HDV — should we edit it “natively” or by using an “intermediate” codec. Be prepared to see this technical debate — which is the subject of much marketing hand-waving and editor hand-wringing — from a different perspective.
In the fall, I will focus specifically on this issue by working with Edius 4 from Canopus/Grass Valley. Edius has the ability edit using native HDV as well as an HD intermediate codec, HQ. Like Liquid, it supports multi-format timelines, including 720p24.
In future installments, I will cover auto-handling different aspect ratios for NTSC tape, widescreen DVD, and HD timelines; Dolby 5.1 sound (a value-added function for DVDs and a “must-have” for high-def DVDs); HDV record-to-tape (what is “Smart GOP” MPEG-2 editing and when does it reduce export time?); plus HD-to-SD DVD production.
